[eszed]

I agree that attestation is the biggest deal. My current hope is that the upcoming GrapheneOS phones will be able to achieve that. It's really up to a manufacturer being able to strong-arm third-parties - banks and such - into accepting their chain of trust, and Motorola may be able to do that.

[Aachen]

GrapheneOS doesn't solve the attestation problem though. If you compile it from source, the attestation still breaks (namely, you have to convince vendors that use attestation to trust your key). In practice, it does not allow you to run your code on your device anymore than stock Android with Google services does. At best, if GrapheneOS' build is reproducible, you can view the source code online and know that there isn't anything else in the blob you load onto your device

Source-available isn't quite the same as having the software freedoms (use, study, modify, share) where you can modify the code or inspect what the various third-party apps are doing on your device. I can currently look into /data/data/any_app and modify preferences, view what telemetry is queued up, remove gigabytes of cache files... all that goes away with a GrapheneOS installation that passes attestation. They want to appear legitimate to app vendors and so comply with Google's rules about what data is accessible to users; otherwise, they'd never convince anyone to add their attestation keys to the allowlist. You need to be on a closed device before those vendors put you on the allow list (and you probably need to sell at least a million devices before they bother to consider you). The concept of attesting your phone is fundamentally antithetical to open source

[microtonal]

I don't think the upcoming Motorola phones will change this by themselves. As far as I understand, Motorola will not directly sell GrapheneOS phones. They will make phones that fulfill the hardware requirements and work with the GrapheneOS team to make the firmware available, etc. It will still be up to the user to install GrapheneOS.

That has benefits - you do not have to trust Motorola not to ship stuff that you wouldn't want in the image. They are pristine images that the GrapheneOS project provides. But it also probably doesn't get Motorola in hot water with Google, since they partially do the same as Google does (provide phones with unlocked bootloaders) and what Google used to do (open drivers, device trees, etc.). Plus there are other OEMs that have similar partnerships with other projects (e.g. Fairphone).

into accepting their chain of trust

It's GrapheneOS who will sign the images, not Motorola.

Speaking of Europe, Motorola is probably not big enough here to strong-arm parties. Besides that, I don't think they will do that, since they have to stay in good graces with Google for distributing GMS Android.

I think for Motorola, there are three wins: 1. GrapheneOS has hundreds of thousands of users now, for a smaller OEM capturing some of that market is a significant addition; 2. they want to have a security-focused offering; GrapheneOS can provide that; and 3. they probably want to strengthen their position towards Google. Samsung has their own app store, device finding ecosystem, etc., this tells Google - if you take too much power, we can go our own way. If the Motorola-GrapheneOS experiment is successful, this could provide a similar contingency plan that might hold Google from trying to reign in OEMs too much. This is a real risk for Google - Pixel is so small in terms of marketshare that if, say Samsung, would go on its own, Google Android is pretty much dead.

At any rate, I think Motorola-GrapheneOS can have more of an indirect effect. I think Play Integrity remote attestation will fall if GrapheneOS can quickly get so many users that they become a force to reckon with. In the past, it helped when GrapheneOS users e-mailed an app developer that switched to strong Play Integrity. This will be much more powerful if the GrapheneOS user base grows 10x and more growth is probably possible if there are non-Google devices (especially because part of the potential user base does not want to give a cent to Google).

Getting the EU to ban Play Integrity as-is probably has a much larger chance of succeeding though. This is why I always recommend people to file a DMA complaint/contact the DMA team when some app gets blocked on alternative ROMs due to Play Integrity. The DMA team needs to see/feel that this affects a lot of real people.

[inigyou]

> Getting the EU to ban Play Integrity

Unlikely to happen, it's just not how they operate. Instead they will create a government registry where any legal entity can register their integrity attestation keys, with oversight by a bureaucracy in a process that takes 6 months and €100,000. Apps will be required to use the registry but still allowed to block attestations they don't like, and any attestation key that leaks into open circulation (such as FOSS) will be blacklisted in the registry because it no longer attests anything.

[microtonal]

any legal entity can register their integrity attestation keys, with oversight by a bureaucracy in a process that takes 6 months and €100,000

That wouldn't be great, but at the same time an improvement over the current situation.

attestation key that leaks into open circulation (such as FOSS) will be blacklisted in the registry because it no longer attests anything

Which makes sense if remote attestation is what you want.

[inigyou]

Yes. It makes sense for well-regulated remote attestation. But is that what we actually want or do we want to destroy the concept altogether?